Using AgentRouter with Codex

This guide will help you configure OpenAI Codex to use AgentRouter as your API provider.

Install Codex

pnpm install -g @openai/codex

npm install -g @openai/codex

yarn global add @openai/codex

Configure Environment Variables

Visit agentrouter.org/console/token to get your API Key
Set the system environment variable AGENT_ROUTER_TOKEN with your API key
Create the configuration file ~/.codex/config.toml:

model = "gpt-5"  # Can also be: glm-4.5, glm-4.6, deepseek-v3.1
model_provider = "openai-chat-completions"
preferred_auth_method = "apikey"

[model_providers.openai-chat-completions]
name = "OpenAI using Chat Completions"
base_url = "https://agentrouter.org/v1"
env_key = "AGENT_ROUTER_TOKEN"
wire_api = "chat"
query_params = {}
stream_idle_timeout_ms = 300000

Create the auth file ~/.codex/auth.json:

{
  "OPENAI_API_KEY": "YOUR_AGENTROUTER_API_KEY"
}

Important

Replace YOUR_AGENTROUTER_API_KEY with your actual API key from AgentRouter Console.

Launch Codex

After configuration, navigate to your project directory and start Codex:

# Navigate to your project directory
cd my-project

# Launch OpenAI Codex
codex

Using Codex in VSCode

Install Codex CLI following the instructions above
Open VSCode and install the Codex extension
Go to Settings and click the icon in the upper right corner to switch to JSON configuration mode

Codex Config Step 1

Add the following configuration:

{
  "chatgpt.apiBase": "https://agentrouter.org/v1",
  "chatgpt.config": {
    "preferred_auth_method": "api_key",
    "model_provider": "openai-chat-completions"
  }
}

Codex Config Step 2

Click the Codex icon to start using it!

Available Models

Model	Description
`gpt-5`	Latest GPT model
`glm-4.5`	GLM 4.5 model
`glm-4.6`	GLM 4.6 model
`deepseek-v3.1`	DeepSeek V3.1 model
`deepseek-v3.2`	DeepSeek V3.2 model

Using AgentRouter with OpenAI Codex